TWENTY-EIGHT PARAPLEGIC, QUADRUPLEGIC, CEREBRAL PALSIED, AND OTHER SEVERELY PHYSICALLY HANDICAPPED SECONDARY STUDENTS, PATIENTS IN THE LOS ANGELES COUNTY RANCHO LOS AMIGOS HOSPITAL, WERE INCLUDED IN A STUDY OF THE EFFECTIVENESS OF AUTOMATED VISUAL PROGRAMED INSTRUCTION. SUBJECTS WERE DIVIDED INTO FOUR MATCHED GROUPS BY READING LEVEL AND INTELLIGENCE. FOUR TREATMENT MODES WERE USED TO TEACH ARITHMETIC FRACTIONS AND DECIMALS. TWO GROUPS ALTERNATED BETWEEN TEACHING MACHINES (TM) AND CLASSROOM (C). ONE GROUP REMAINED CONTINUOUSLY WITH THE TEACHER AND ONE CONTINUOUSLY WITH THE MACHINES. MACHINE CONTROLS WERE ADAPTED TO DISABILITIES. INDEPENDENT VARIABLES WERE--(1) INSTRUCTION MATERIALS WITH UNITS SPLIT AT MIDPOINT PROVIDING FOUR UNITS--INSTRUCTIONAL CONTENT USED BY MACHINES AND TEACHERS WAS IDENTICAL, (2) MODES OF INSTRUCTION, AND (3) MATCHING CRITERIA, READING LEVEL AND INTELLIGENCE. DEPENDENT VARIABLES WERE--(1) TIME REQUIRED TO COMPLETE UNITS, (2) PERFORMANCE IN TERMS OF MEAN DIFFERENCE SCORES (PRETEST, MIDTEST, POST-TEST), AND (3) RATE OF LEARNING AND PERCENT OF ERROR. FINDINGS WERE--(1) THE TM MODE OF INSTRUCTION WAS ABOUT TWO-THIRDS MORE EFFICIENT IN TIME, (2) THE TM MODE WAS MOST EFFECTIVE IN TANDEM WITH THE C MODE, (3) THE TM MODE WAS MOST EFFECTIVE WITH LOWER INTELLIGENCE SUBJECTS, (4) THE C INSTRUCTION MODE BECAME MORE EFFECTIVE AS INSTRUCTION MATERIAL BECAME MORE COMPLEX (DIFFICULT), (5) THE MOST EFFECTIVE SEQUENCE IS TM FOLLOWED BY C INSTRUCTION, AND (6) OPERATION OF MACHINES CAN BE ADAPTED TO DISABILITIES. (TC)